I'm analyzing some ontologies with the pellet reasoner, but I'm getting some strange results. When I perform the "pellet info" method on an ontology, and compare the OWL Profile to the DL Expressivity, I sometimes get ontologies that have different expressivities bu the same profile. That seems at least plausible, but then I am also finding ontologies with the same expressivity but different OWL Profiles. How is this possible, since the profile is a mark of the ontology's expressivity and reducibility in the first place?
views:
25answers:
1
+1
A:
If you're considering comparing ontologies under OWL2 (based on SROIQ) in general and, for example, a profile which is a sub-language such as OWL-EL (based on EL++), then it is conceivable that an ontology can have the same expressivity and different 'profile', if you consider full OWL2 to be a 'profile' and expressivity to be DL-expressivity, since SROIQ subsumes EL++.
However, if by profile you strictly mean one of the three distinct OWL2 profiles; OWL2-EL, OWL2-QL and OWL2-RL, then it is still possible that two ontologies with different OWL2 profiles can indeed have the same expressivity (at least, DL-expressivity) as these profiles are underpinned by description logics that may have some intersecting expressivity.
For example, ontologies in the OWL2-EL profile (based on EL++) and in the OWL2-QL profile (based on DL-Lite), both of these description logics can express things such as:
- Concept inclusion axioms
- Instance assertions against concepts and roles
- Existential property restrictions
- Conjunctive concept expressions
Ontologies in either profile (EL, QL) which are restricted to usage of the DL constructs which are common to both would therefore have the same DL-expressivity.
sharky
2010-09-08 07:14:49